[ { "memory": 1493712896, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 1, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1501642752, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 1, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1509572608, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 1, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1510260736, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 1, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1510948864, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 1, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1513308160, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 1, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1521238016, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 1, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1523597312, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 1, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1525956608, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 1, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1528315904, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 1, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1530675200, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 1, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1531363328, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 1, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1539293184, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 1, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1547223040, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 1, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1555152896, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 1, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1555841024, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 1, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1558200320, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 1, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1558888448, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 1, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1559576576, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 1, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1561935872, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 1, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1569865728, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 1, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1572225024, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 1, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1572913152, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 1, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1580843008, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 1, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1588772864, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 1, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1589460992, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 1, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1590149120, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 1, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1598078976, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 1, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1600438272, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 1, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1608368128, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 1, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1610727424, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 1, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1618657280, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 1, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1619345408, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 1, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1620033536, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 1, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1627963392, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 1, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1635893248, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 1, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1643823104, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 1, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1646182400, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 1, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1648541696, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 1, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1650900992, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 1, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1653260288, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 1, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1655619584, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 1, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1663549440, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 1, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1671479296, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 1, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1679409152, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 1, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1680097280, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 1, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1680785408, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 1, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1681473536, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 1, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1682161664, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 1, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1682849792, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 1, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1690779648, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 1, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1691467776, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 1, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1692155904, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 1, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1692844032, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 1, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1700773888, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 1, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1701462016, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 1, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1709391872, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 1, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1711751168, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 1, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1719681024, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 1, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1727610880, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 1, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1728299008, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 1, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1730658304, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 1, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1733017600, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 1, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1740947456, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 1, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1741635584, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 1, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1749565440, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 1, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1751924736, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 1, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1752612864, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 1, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1754972160, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 1, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1762902016, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 1, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1770831872, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 1, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1771520000, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 1, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1779449856, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 1, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1787379712, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 1, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1795309568, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 1, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1803239424, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 1, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1805598720, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 1, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1807958016, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 1, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1815887872, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 1, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1818247168, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 1, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1818935296, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 1, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1821294592, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 1, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1821982720, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 1, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1822670848, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 1, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1825030144, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 1, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1827389440, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 1, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1828077568, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 1, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1830436864, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 1, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1832796160, "layers": { "layer.0.self_attn.q_proj": 1, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1835155456, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 1, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1835843584, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 1, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1843773440, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 1, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1844461568, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 1, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1845149696, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 1, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1845837824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 1, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1848197120, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 1, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1850556416, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 1, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1851244544, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 1, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1853603840, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 1, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1854291968, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 1, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1862221824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 1, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1864581120, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 1, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1866940416, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 1, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1874870272, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 1, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1875558400, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 1, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1877917696, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 1, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1880276992, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 1, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1882636288, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 1, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1883324416, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 1, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1885683712, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 1, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1886371840, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 1, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1887059968, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 1, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1894989824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 1, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1902919680, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 1, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1903607808, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 1, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1905967104, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 1, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1908326400, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 1, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1910685696, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 1, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1911373824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 1, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1919303680, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 1, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1921662976, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 1, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1929592832, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 1, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1937522688, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 1, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1945452544, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 1, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1953382400, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 1, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1955741696, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 1, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1963671552, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 1, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1964359680, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 1, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1966718976, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 1, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1974648832, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 1, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1977008128, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 1, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1979367424, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 1, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1987297280, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 1, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1987985408, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 1, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1988673536, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 1, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1991032832, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 1, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1991720960, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 1, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1994080256, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 1, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1994768384, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 1, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1995456512, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 1, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 1996144640, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 1, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 2004074496, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 1, "layer.31.mlp.down_proj": 1 } }, { "memory": 2012004352, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 1, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2019934208, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 1, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2020622336, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 1, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2021310464, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 1, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2029240320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 1, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2037170176, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 1, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2045100032, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 1, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2053029888, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 1, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2055389184, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 1, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2057748480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 1, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2058436608, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 1, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2066366464, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 1, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2068725760, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 1, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2076655616, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 1, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2077343744, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 1, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2078031872, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 1, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2085961728, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 1, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2093891584, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 1, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2096250880, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 1, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2104180736, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 1, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2112110592, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 1, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2120040448, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 1, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2127970304, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 1, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2130329600, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 1, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2138259456, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 1, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2146189312, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 1, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2154119168, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 1, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2162049024, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 1, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2164408320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 1, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2165096448, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 1, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2167455744, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 1, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2168143872, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 1, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2170503168, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 1, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2178433024, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 1, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2186362880, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 1, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2187051008, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 1, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2194980864, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 1, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2197340160, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 1, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2205270016, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 1, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2213199872, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 1, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2213888000, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 1, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2221817856, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 1, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2229747712, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 1, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2237677568, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 1, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2245607424, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 1, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2246295552, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 1, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2246983680, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 1, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2247671808, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 1, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2255601664, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 1, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2263531520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 1, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2265890816, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 1, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2273820672, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 1, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2276179968, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 1, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2284109824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 1, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2292039680, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 1, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2292727808, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 1, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2300657664, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 1, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2308587520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 1, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2316517376, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 1, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2324447232, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 1, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2332377088, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 1, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2340306944, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 1, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2348236800, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 1, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2356166656, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 1, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2364096512, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 1, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2366455808, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 1, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2374385664, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 1, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2382315520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 1, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2390245376, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 1, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2398175232, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 1 } }, { "memory": 2406105088, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 1, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2414034944, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 1, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2416394240, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 1, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2424324096, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 1, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2425012224, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 1, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2432942080, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 1, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2433630208, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 1, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2435989504, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 1, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2438348800, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 1, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2446278656, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 1, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2446966784, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 1, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2449326080, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 1, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2450014208, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 2, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2457944064, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 2, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2465873920, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 2, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2473803776, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 2, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2476163072, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 2, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2484092928, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 2, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2492022784, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 2, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2499952640, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 2, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2507882496, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 2, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2515812352, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 2, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2518171648, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 2, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2526101504, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 2, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2534031360, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 2, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2541961216, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 2, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2549891072, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 2, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2550579200, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 2, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2551267328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 2, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2559197184, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 2, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2559885312, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 2, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2562244608, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 2, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2564603904, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 2, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2572533760, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 2, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2574893056, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 2, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2582822912, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 2, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2585182208, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 2, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2593112064, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 2, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2595471360, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 2, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2597830656, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 2, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2598518784, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 2, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2600878080, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 2, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2608807936, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 2, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2616737792, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 2, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2624667648, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 2, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2625355776, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 2, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2633285632, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 2, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2635644928, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 2, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2643574784, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 2, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2651504640, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 2, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2652192768, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 2, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2660122624, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 2, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2668052480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 2, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2670411776, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 2, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2672771072, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 2, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2675130368, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 2, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2683060224, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 2, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2685419520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 2, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2693349376, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 2, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2701279232, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 2, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2701967360, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 2, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2702655488, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 2, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2703343616, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 2, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2711273472, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 2, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2719203328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 2, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2721562624, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 2, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2729492480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 2, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2737422336, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 2, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2745352192, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 2, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2746040320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 2, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2753970176, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 2, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2761900032, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 2, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2769829888, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 2, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2777759744, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 2, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2785689600, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 2, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2786377728, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 2, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2787065856, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 2, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2787753984, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 2, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2795683840, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 2, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2803613696, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 2, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2804301824, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 2, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2804989952, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 2, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2812919808, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 2, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2820849664, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 2, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2828779520, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 2, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2836709376, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 2, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2837397504, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 2, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2839756800, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 2, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2847686656, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 2, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2855616512, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 2, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2856304640, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 2, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2856992768, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 2, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2864922624, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 2, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2872852480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 2, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2873540608, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 2, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2874228736, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 2, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2882158592, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 2, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2884517888, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 2, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2886877184, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 2, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2889236480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 2, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2897166336, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 2, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2897854464, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 2, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2905784320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 2, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2913714176, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 2, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2916073472, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 2, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2924003328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 2, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2931933184, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 2, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2934292480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 2, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2942222336, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 2, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2944581632, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 2, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2952511488, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 2, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2960441344, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 2, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2968371200, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 2, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2976301056, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 2, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2976989184, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 2, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2979348480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 2, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2987278336, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 2, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2995208192, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 2, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 2995896320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 2, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3003826176, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 2, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3004514304, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 2, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3012444160, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 2, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3014803456, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 2, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3022733312, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 2, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3030663168, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 2, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3038593024, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 2, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3039281152, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 2, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3041640448, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 2, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3043999744, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 2, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3044687872, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 2, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3047047168, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 2, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3049406464, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 2, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3051765760, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 2, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3059695616, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 2, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3062054912, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 2, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3069984768, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 2, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3077914624, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 2, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3085844480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 2, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3093774336, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 2, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3094462464, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 2, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3102392320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 2, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3110322176, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 2, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3118252032, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 2, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3120611328, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 2, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3122970624, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 2, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3125329920, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 2, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3133259776, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 2, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3135619072, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 2, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3136307200, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 2, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3138666496, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 2, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3146596352, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 2, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3147284480, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 2, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3149643776, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 2, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3150331904, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 2, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3158261760, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 2 } }, { "memory": 3166191616, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 2, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3166879744, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 2, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3174809600, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 2, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3177168896, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 2, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3177857024, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 2, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3180216320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 2, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3180904448, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 2, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3181592576, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 2, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3183951872, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 2, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3186311168, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 2, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3186999296, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 2, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3189358592, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 2, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3197288448, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 2, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3205218304, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 2, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3207577600, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 2, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3209936896, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 2, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3212296192, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 2, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3220226048, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 2, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3220914176, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 2, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3221602304, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 2, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3222290432, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 2, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3222978560, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 2, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3230908416, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 2, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3233267712, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 2, "layer.31.mlp.down_proj": 3 } }, { "memory": 3241197568, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 2, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3249127424, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 2, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3249815552, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 2, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3250503680, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 2, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3252862976, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 2, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3255222272, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 2, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3263152128, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 2, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3265511424, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 2, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3267870720, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 2, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3270230016, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 2, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3272589312, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 2, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3280519168, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 2, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3281207296, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 2, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3289137152, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 2, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3291496448, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 2, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3293855744, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 2, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3294543872, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 2, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3296903168, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 2, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3299262464, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 2, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3307192320, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 2, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3315122176, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 2, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3315810304, "layers": { "layer.0.self_attn.q_proj": 2, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3318169600, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 2, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3318857728, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 2, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3326787584, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 2, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3334717440, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 2, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3335405568, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 2, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3336093696, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 2, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3344023552, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 2, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3344711680, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 2, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3345399808, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 2, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3347759104, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 2, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3348447232, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 2, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3349135360, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 2, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3349823488, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 2, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3357753344, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 2, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3360112640, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 2, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3360800768, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 2, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3361488896, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 2, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3363848192, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 2, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3364536320, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 2, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3372466176, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 2, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3374825472, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 2, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3375513600, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 2, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3376201728, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 2, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3376889856, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 2, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3377577984, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 2, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3378266112, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 2, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3378954240, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 2, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3386884096, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 2, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3389243392, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 2, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3389931520, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 2, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3392290816, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 2, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3392978944, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 2, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3400908800, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 2, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3401596928, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 2, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3403956224, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 2, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3406315520, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 3, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3408674816, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 3, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3416604672, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 3, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3424534528, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 3, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3432464384, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 3, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3440394240, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 3, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3441082368, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 3, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3449012224, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 3, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3451371520, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 3, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3459301376, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 3, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3467231232, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 3, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3475161088, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 3, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3475849216, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 3, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3483779072, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 3, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3491708928, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 3, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3499638784, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 3, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3507568640, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 3, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3515498496, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 3, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3516186624, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 3, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3524116480, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 3, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3532046336, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 3, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3539976192, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 3, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3540664320, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 3, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3548594176, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 3, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3550953472, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 3, "layer.31.mlp.down_proj": 3 } }, { "memory": 3558883328, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 3, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3559571456, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 3, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3567501312, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 3, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3569860608, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 3, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3577790464, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 3, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3585720320, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 3, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3593650176, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 3, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3601580032, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 3, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3609509888, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 3, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3617439744, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 3, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3625369600, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 3, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3633299456, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 3, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3635658752, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 3, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3643588608, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 3, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3645947904, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 3, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3648307200, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 3, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3650666496, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 3, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3658596352, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 3, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3666526208, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 3, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3674456064, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 3, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3682385920, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 3, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3690315776, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 3, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3692675072, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 3, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3700604928, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 3 } }, { "memory": 3708534784, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 3, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3709222912, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 3, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3717152768, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 3, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3725082624, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 3, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3727441920, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 3, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3735371776, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 3, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3743301632, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 3, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3751231488, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 3, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3759161344, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 3, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3767091200, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 3, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3775021056, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 3, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3775709184, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 3, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3783639040, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 3, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3791568896, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 3, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3799498752, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 3, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3807428608, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 3, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3815358464, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 3, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3823288320, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 3, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3823976448, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 3, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3831906304, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 3, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3839836160, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 3, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3847766016, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 3, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3855695872, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 3, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3863625728, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 3, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3871555584, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 3, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3879485440, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 3, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3880173568, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 3, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3888103424, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 3, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3896033280, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 3, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3903963136, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 3, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3911892992, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 3, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3919822848, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 3, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3927752704, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 3, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3935682560, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 3, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3943612416, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 3, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3945971712, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 3, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3948331008, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 3, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3956260864, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 3, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3956948992, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 3, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3957637120, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 3, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3958325248, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 3, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3966255104, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 3, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3974184960, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 3, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3976544256, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 3, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3984474112, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 3, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3985162240, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 3, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3987521536, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 3, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 3995451392, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 3, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4003381248, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 3, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4005740544, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 3, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4008099840, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 3, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4016029696, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 3, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4018388992, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 3, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4026318848, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 3, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4034248704, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 3, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4042178560, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 3, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4044537856, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 3, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4052467712, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 3, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4060397568, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 3, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4061085696, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 3, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4069015552, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 3, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4071374848, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 3, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4072062976, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 3, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4079992832, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 3, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4087922688, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 3, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4088610816, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 3, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4096540672, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 3, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4098899968, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 3, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4099588096, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 3, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4101947392, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 3, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4104306688, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 3, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4106665984, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 3, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4114595840, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 3, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4122525696, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 3, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4130455552, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 3, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4138385408, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 3, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4139073536, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 3, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4139761664, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 3, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4147691520, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 3, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4155621376, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 3, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4156309504, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 3, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4164239360, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 3, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4166598656, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 3, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4174528512, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 3, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4182458368, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 3, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4183146496, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 3, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4191076352, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 3, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4193435648, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 3, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4195794944, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 3, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4196483072, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 3, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4204412928, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 3, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4205101056, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 3, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4205789184, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 3, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4206477312, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 3, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4214407168, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 3, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4222337024, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 3, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4223025152, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 3, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4225384448, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 3, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4227743744, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 3, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4228431872, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 3, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4230791168, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 3, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4233150464, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 3, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4233838592, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 3, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4236197888, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 3, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4238557184, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 3, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4240916480, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 3, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4243275776, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 3, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4245635072, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 3, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4247994368, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 3, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4250353664, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 3, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4251041792, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 3, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4253401088, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 3, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4261330944, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 3, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4263690240, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 3, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4264378368, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 3, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4265066496, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 3, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4267425792, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 3, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4268113920, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 3, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4270473216, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 3, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4272832512, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 3, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4273520640, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 3, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4275879936, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 3, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4276568064, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 3, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4278927360, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 3, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4279615488, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 3, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4281974784, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 3, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4282662912, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 3, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4283351040, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 3, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4285710336, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 3, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4286398464, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 3, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4288757760, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 3, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4289445888, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 3, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4290134016, "layers": { "layer.0.self_attn.q_proj": 3, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4292493312, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 3, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4293181440, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 3, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4301111296, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 3, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4301799424, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 3, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4302487552, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 3, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4303175680, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 3, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4303863808, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 3, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4306223104, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 3, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4306911232, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 3, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4309270528, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 3, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4309958656, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 3, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4312317952, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 3, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4313006080, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 3, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4315365376, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 3, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4316053504, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 3, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4316741632, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 3, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4319100928, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 3, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4319789056, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 3, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4322148352, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 3, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4324507648, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 3, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4332437504, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 3, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4333125632, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 3, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4333813760, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 3, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4334501888, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 3, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4335190016, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 3, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4337549312, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 3, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4339908608, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 3, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4342267904, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 3, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4344627200, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 3, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4346986496, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 3, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4347674624, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 3, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4348362752, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 3, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4349050880, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 3, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4351410176, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 3, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4352098304, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 3, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4352786432, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 3, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4355145728, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 3, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4355833856, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 3, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4356521984, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 3, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4358881280, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 3, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4359569408, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 3, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4361928704, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 3, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4362616832, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 4, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4364976128, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 4, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4372905984, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 4, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4380835840, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 4, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4388765696, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 4, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4396695552, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 4, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4399054848, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 4, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4406984704, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 4, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4414914560, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 4, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4422844416, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 4, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4430774272, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 4, "layer.31.mlp.down_proj": 4 } }, { "memory": 4438704128, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 4, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 4 } }, { "memory": 4446633984, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 4, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 4 } }, { "memory": 4454563840, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 4, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 4 } }, { "memory": 4456923136, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 4 } }, { "memory": 4464852992, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 4, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4467212288, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 4, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4467900416, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 4, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4468588544, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 4, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4476518400, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 4, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4484448256, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 4, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4492378112, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 4, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4494737408, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 4, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4502667264, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 4, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4510597120, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 4, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4518526976, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 4, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4526456832, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 4, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4534386688, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 4, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4535074816, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 4, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4543004672, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 4, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4550934528, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 4, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4558864384, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 4, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4566794240, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 4, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4574724096, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 4, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4582653952, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 4, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4590583808, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 4, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4598513664, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 4, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4606443520, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 4, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4614373376, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 4, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4622303232, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 4, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4630233088, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 4, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4638162944, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 4, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4638851072, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 4, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4646780928, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 4, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4654710784, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 4, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4655398912, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 4, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4663328768, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 4, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4671258624, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 4, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4679188480, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 4, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4679876608, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 4, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4682235904, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 4, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4690165760, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 4, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4690853888, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 4, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4698783744, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 4, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4706713600, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 4, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4714643456, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 4, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4722573312, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 4, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4730503168, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 4, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4738433024, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 4, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4746362880, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 4, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4754292736, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 4, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4756652032, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 4, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4764581888, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 4, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4766941184, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 4, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4774871040, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 4, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4775559168, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 4, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4776247296, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 4, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4784177152, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 4, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4784865280, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 4, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4792795136, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 4, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4800724992, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 4, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4803084288, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 4, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4805443584, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 4, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4813373440, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 4, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4821303296, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 4, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4821991424, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 4, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4829921280, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 4, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4832280576, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 4, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4832968704, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 4, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4840898560, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 4, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4843257856, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 4, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4851187712, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 4, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4851875840, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 4, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4859805696, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 4, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4867735552, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 4, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4875665408, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 4, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4883595264, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 4, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4891525120, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 4, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4899454976, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 4, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4907384832, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 4, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4915314688, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 4, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4923244544, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 4, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4923932672, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 4, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4926291968, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 4, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4934221824, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 4, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4936581120, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 4, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4938940416, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 4, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4946870272, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 4, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4949229568, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 4, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4957159424, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 4, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4957847552, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 4, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4965777408, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 4, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4973707264, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 4, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4981637120, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 4, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4982325248, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 4, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4990255104, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 4, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 4992614400, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 4, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5000544256, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 4, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5001232384, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 4, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5003591680, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 4, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5005950976, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 4, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5013880832, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 4, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5016240128, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 4, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5024169984, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 4, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5024858112, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 4, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5025546240, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 4, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5033476096, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 4, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5035835392, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 4, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5038194688, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 4, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5046124544, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 4, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5054054400, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 4, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5056413696, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 4, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5064343552, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 4, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5066702848, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 4, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5074632704, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 4, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5082562560, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 4, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5083250688, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 4, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5083938816, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 4, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5091868672, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 4, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5092556800, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 4, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5094916096, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 4, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5095604224, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 4, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5103534080, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 4, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5105893376, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 4, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5108252672, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 4, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5108940800, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 4, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5111300096, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 4, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5111988224, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 4, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5119918080, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 4, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5127847936, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 4, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5135777792, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 4, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5138137088, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 4, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5138825216, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 4, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5141184512, "layers": { "layer.0.self_attn.q_proj": 4, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5143543808, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 4, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5145903104, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 4, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5148262400, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 4, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5148950528, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 4, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5149638656, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 4, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5150326784, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 4, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5152686080, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 4, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5155045376, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 4, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5155733504, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 4, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5158092800, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 4, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5158780928, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 4, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5159469056, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 4, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5161828352, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 4, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5164187648, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 4, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5166546944, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 4, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5167235072, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 4, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5169594368, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 4, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5171953664, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 4, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5172641792, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 4, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5173329920, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 4, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5174018048, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 4, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5174706176, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 4, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5175394304, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 4, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5176082432, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 4, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5178441728, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 4, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5179129856, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 4, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5181489152, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 4, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5182177280, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 4, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5190107136, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 4, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5198036992, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 4, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5200396288, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 4, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5201084416, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 4, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5203443712, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 4, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5204131840, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 4, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5206491136, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 4, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5207179264, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 4, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5215109120, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 4, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5217468416, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 4, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5219827712, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 4, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5227757568, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 4, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5228445696, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 4, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5229133824, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 4, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5229821952, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 4, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5232181248, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 4, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5232869376, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 4, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5240799232, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 4, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5243158528, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 4, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5245517824, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 4, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5247877120, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 4, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5255806976, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 4, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5258166272, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 4, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5260525568, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 4, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5261213696, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 4, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5261901824, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 4, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5264261120, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 4, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5264949248, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 4, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5267308544, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 4, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5267996672, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 4, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5268684800, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 4, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5269372928, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 4, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5277302784, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 4, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5277990912, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 4, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5280350208, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 4, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5281038336, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 4, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5283397632, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 4, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5285756928, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 4, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5288116224, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 4, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5290475520, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 4, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5291163648, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 4, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5291851776, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 4, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5292539904, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 4, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5300469760, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 4, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5301157888, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 4, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5303517184, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 4, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5305876480, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 4, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5308235776, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 4, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5316165632, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 4, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5316853760, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 4, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5317541888, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 4, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5318230016, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 4, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5318918144, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 5, "layer.31.mlp.down_proj": 5 } }, { "memory": 5326848000, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 5, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 5 } }, { "memory": 5334777856, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 5, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 5 } }, { "memory": 5342707712, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 5, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 5 } }, { "memory": 5350637568, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 5, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 5 } }, { "memory": 5358567424, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 5, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 5 } }, { "memory": 5366497280, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 5, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 5 } }, { "memory": 5368856576, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 5 } }, { "memory": 5376786432, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 5, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5384716288, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 5, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5392646144, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 5, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5395005440, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 5, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5402935296, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 5, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5410865152, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 5, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5418795008, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 5, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5426724864, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 5, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5434654720, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 5, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5442584576, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 5, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5450514432, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 5, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5458444288, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 5, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5466374144, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 5, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5474304000, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 5, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5482233856, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 5, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5490163712, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 5, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5498093568, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 5, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5506023424, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 5, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5513953280, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 5, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5521883136, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 5, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5529812992, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 5, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5532172288, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 5, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5540102144, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 5, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5540790272, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 5, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5548720128, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 5, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5556649984, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 5, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5564579840, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 5, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5572509696, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 5, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5580439552, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 5, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5581127680, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 5, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5589057536, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 5, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5596987392, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 5, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5604917248, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 5, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5612847104, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 5, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5615206400, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 5, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5623136256, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 5, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5631066112, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 5, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5633425408, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 5, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5641355264, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 5, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5649285120, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 5, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5657214976, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 5, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5657903104, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 5, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5665832960, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 5, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5673762816, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 5, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5681692672, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 5, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5689622528, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 5, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5697552384, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 5, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5699911680, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 5, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5707841536, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 5, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5710200832, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 5, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5718130688, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 5, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5720489984, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 5, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5728419840, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 5, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5729107968, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 5, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5737037824, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 5, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5744967680, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 5, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5752897536, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 5, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5755256832, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 5, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5763186688, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 5, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5763874816, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 5, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5766234112, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 5, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5766922240, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 5, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5774852096, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 5, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5775540224, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 5, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5783470080, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 5, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5785829376, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 5, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5788188672, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 5, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5788876800, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 5, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5796806656, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 5, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5799165952, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 5, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5799854080, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 5, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5800542208, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 5, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5808472064, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 5, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5810831360, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 5, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5818761216, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 5, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5826691072, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 5, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5834620928, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 5, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5842550784, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 5, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5850480640, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 5, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5858410496, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 5, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5866340352, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 5, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5867028480, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 5, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5874958336, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 5, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5882888192, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 5, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5885247488, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 5, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5885935616, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 5, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5886623744, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 5, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5888983040, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 5, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5891342336, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 5, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5899272192, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 5, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5907202048, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 5, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5909561344, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 5, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5910249472, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 5, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5918179328, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 5, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5918867456, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 5, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5921226752, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 5, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5929156608, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 5, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5931515904, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 5, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5932204032, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 5, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5940133888, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 5, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5942493184, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 5, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5950423040, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 5, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5951111168, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 5, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5953470464, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 5, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5954158592, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 5, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5956517888, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 5, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5957206016, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 5, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5965135872, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 5, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5967495168, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 5, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5975425024, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 5, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5976113152, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 5, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5978472448, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 5, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5986402304, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 5, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 5994332160, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 5, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6002262016, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 5, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6004621312, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 5, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6006980608, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 5, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6014910464, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 5, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6015598592, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 5, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6017957888, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 5, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6018646016, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 5, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6021005312, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 5, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6028935168, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 5, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6031294464, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 5, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6039224320, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 5, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6039912448, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 5, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6042271744, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 5, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6050201600, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 5, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6058131456, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 5, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6060490752, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 5, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6061178880, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 5, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6061867008, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 5, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6064226304, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 5, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6064914432, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 5, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6067273728, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 5, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6067961856, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 5, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6075891712, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 5, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6076579840, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 5, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6077267968, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 5, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6077956096, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 5, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6085885952, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 5, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6088245248, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 5, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6090604544, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 5, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6092963840, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 5, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6093651968, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 5, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6096011264, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 5, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6096699392, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 5, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6097387520, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 5, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6098075648, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 5, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6100434944, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 5, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6108364800, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 5, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6109052928, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 5, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6109741056, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 5, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6117670912, "layers": { "layer.0.self_attn.q_proj": 5, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6120030208, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 5, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6127960064, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 5, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6130319360, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 5, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6131007488, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 5, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6133366784, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 5, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6135726080, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 5, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6136414208, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 5, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6144344064, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 5, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6152273920, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 5, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6154633216, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 5, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6162563072, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 5, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6164922368, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 5, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6165610496, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 5, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6173540352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 5, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6175899648, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 5, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6176587776, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 5, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6177275904, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 5, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6185205760, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 5, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6187565056, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 5, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6189924352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 5, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6190612480, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 5, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6191300608, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 5, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6191988736, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 5, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6192676864, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 5, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6193364992, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 5, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6195724288, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 5, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6196412416, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 5, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6198771712, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 5, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6201131008, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 5, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6201819136, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 5, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6202507264, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 5, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6204866560, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 5, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6205554688, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 5, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6207913984, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 5, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6208602112, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 5, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6209290240, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 5, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6217220096, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 5, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6219579392, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 5, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6221938688, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 5, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6224297984, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 5, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6232227840, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 5, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6232915968, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 5, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6240845824, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 5, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6241533952, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 5, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6243893248, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 5, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6244581376, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 5, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6246940672, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 5, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6249299968, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 5, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6249988096, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 5, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6252347392, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 5, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6253035520, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 5, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6253723648, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 5, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6256082944, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 5, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6264012800, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 5, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6264700928, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 5, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6267060224, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 5, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6267748352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 5, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6270107648, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 5, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6272466944, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 5, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6273155072, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 5, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6273843200, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 5, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6274531328, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 5, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6275219456, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 6 } }, { "memory": 6283149312, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 6, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 6291079168, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 6, "layer.31.mlp.down_proj": 7 } }, { "memory": 6299009024, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 6, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6306938880, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 6, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6314868736, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 6, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6322798592, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 6, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6330728448, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 6, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6338658304, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 6, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6346588160, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 6, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6354518016, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 6, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6355206144, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 6, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6363136000, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 6, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6371065856, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 6, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6378995712, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 6, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6386925568, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 6, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6394855424, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 6, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6402785280, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 6, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6405144576, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 6, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6407503872, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 6, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6415433728, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 6, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6423363584, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 6, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6431293440, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 6, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6439223296, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 6, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6447153152, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 6, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6455083008, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 6, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6463012864, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 6, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6470942720, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 6, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6471630848, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 6, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6479560704, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 6, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6487490560, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 6, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6495420416, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 6, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6503350272, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 6, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6511280128, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 6, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6519209984, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 6, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6527139840, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 6, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6535069696, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 6, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6542999552, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 6, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6545358848, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 6, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6546046976, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 6, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6553976832, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 6, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6561906688, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 6, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6569836544, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 6, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6577766400, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 6, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6585696256, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 6, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6593626112, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 6, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6594314240, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 6, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6602244096, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 6, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6602932224, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 6, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6610862080, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 6, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6618791936, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 6, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6626721792, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 6, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6634651648, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 6, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6642581504, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 6, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6650511360, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 6, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6658441216, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 6, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6666371072, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 6, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6668730368, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 6, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6676660224, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 6, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6684590080, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 6, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6692519936, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 6, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6693208064, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 6, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6693896192, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 6, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6701826048, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 6, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6704185344, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 6, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6712115200, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 6, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6712803328, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 6, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6720733184, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 6, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6723092480, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 6, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6731022336, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 6, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6738952192, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 6, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6746882048, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 6, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6749241344, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 6, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6757171200, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 6, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6765101056, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 6, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6767460352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 6, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6775390208, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 6, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6777749504, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 6, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6780108800, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 6, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6788038656, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 6, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6795968512, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 6, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6796656640, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 6, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6804586496, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 6, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6806945792, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 6, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6814875648, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 6, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6815563776, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 6, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6817923072, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 6, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6820282368, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 6, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6828212224, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 6, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6828900352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 6, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6831259648, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 6, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6839189504, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 6, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6839877632, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 6, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6840565760, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 6, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6842925056, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 6, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6845284352, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 6, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6845972480, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 6, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6848331776, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 6, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6856261632, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 6, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6864191488, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 6, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6864879616, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 6, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6865567744, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 6, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6866255872, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 6, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6868615168, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 6, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6869303296, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 6, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6871662592, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 6, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6872350720, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 6, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6880280576, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 6, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6880968704, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 6, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6881656832, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 6, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6889586688, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 6, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6890274816, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 6, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6890962944, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 6, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6898892800, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 6, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6901252096, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 6, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6909181952, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 6, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6911541248, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 6, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6913900544, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 6, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6914588672, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 6, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6916947968, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 6, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6919307264, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 6, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6921666560, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 6, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6924025856, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 6, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6924713984, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 6, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6927073280, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 6, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6935003136, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 6, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6937362432, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 6, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6939721728, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 6, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6940409856, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 6, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6942769152, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 6, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6945128448, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 6, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6947487744, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 6, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6949847040, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 6, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6952206336, "layers": { "layer.0.self_attn.q_proj": 6, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6954565632, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 6, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6956924928, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 6, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6959284224, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 6, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6967214080, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 6, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6967902208, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 6, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6970261504, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 6, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6978191360, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 6, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6986121216, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 6, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6986809344, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 6, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 6994739200, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 6, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7002669056, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 6, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7010598912, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 6, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7011287040, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 6, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7011975168, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 6, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7019905024, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 6, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7022264320, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 6, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7022952448, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 6, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7023640576, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 6, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7024328704, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 6, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7026688000, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 6, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7029047296, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 6, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7031406592, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 6, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7039336448, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 6, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7040024576, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 6, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7047954432, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 6, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7055884288, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 6, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7058243584, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 6, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7058931712, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 6, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7059619840, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 6, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7061979136, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 6, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7069908992, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 6, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7070597120, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 6, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7078526976, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 6, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7086456832, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 6, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7087144960, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 6, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7087833088, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 6, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7090192384, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 6, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7092551680, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 6, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7093239808, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 6, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7093927936, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 6, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7101857792, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 6, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7104217088, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 6, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7104905216, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 6, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7107264512, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 6, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7109623808, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 6, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7117553664, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 6, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7119912960, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 6, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7127842816, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 6, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7130202112, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 6, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7132561408, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 6, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7134920704, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 6, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7135608832, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 6, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7143538688, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 6, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7144226816, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 6, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7144914944, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 6, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7145603072, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 6, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7146291200, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 6, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7146979328, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 6, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7147667456, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 6, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7148355584, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 6, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7150714880, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 6, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7151403008, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 6, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7159332864, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 6, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7161692160, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 6, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7162380288, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 6, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7163068416, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 6, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7163756544, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 6, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7166115840, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 6, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7168475136, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 6, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7169163264, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 6, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7169851392, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 6, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7172210688, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 6, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7172898816, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 6, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7173586944, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 6, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7175946240, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 6, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7176634368, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 6, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7178993664, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 6, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7186923520, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 6, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7189282816, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 6, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7189970944, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 6, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7197900800, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 6, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7200260096, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 6, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7202619392, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 6, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7210549248, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 6, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7211237376, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 6, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7211925504, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 6, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7214284800, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 6, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7222214656, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 6, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7222902784, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 6, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7223590912, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 6, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7231520768, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 7, "layer.31.mlp.down_proj": 7 } }, { "memory": 7239450624, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 7, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7247380480, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 7, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7255310336, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 7, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7257669632, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 7, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7265599488, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 7, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7273529344, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 7, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7281459200, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 7, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7289389056, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 7, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7297318912, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 7, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7305248768, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 7, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7313178624, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 7, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7315537920, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 7, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7323467776, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 7, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7331397632, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 7, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7339327488, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 7 } }, { "memory": 7347257344, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 7, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7355187200, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 7, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7363117056, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 7, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7371046912, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 7, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7378976768, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 7, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7386906624, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 7, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7394836480, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 7, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7402766336, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 7, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7410696192, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 7, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7418626048, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 7, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7420985344, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 7, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7428915200, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 7, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7436845056, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 7, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7444774912, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 7, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7452704768, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 7, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7460634624, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 7, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7468564480, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 7, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7470923776, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 7, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7478853632, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 7, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7486783488, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 7, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7489142784, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 7, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7489830912, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 7, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7497760768, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 7, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7498448896, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 7, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7506378752, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 7, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7514308608, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 7, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7522238464, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 7, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7530168320, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 7, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7538098176, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 7, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7538786304, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 7, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7546716160, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 7, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7554646016, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 7, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7557005312, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 7, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7559364608, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 7, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7561723904, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 7, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7569653760, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 7, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7577583616, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 7, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7585513472, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 7, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7586201600, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 7, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7588560896, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 7, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7596490752, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 7, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7597178880, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 7, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7605108736, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 7, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7613038592, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 7, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7613726720, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 7, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7621656576, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 7, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7624015872, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 7, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7631945728, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 7, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7639875584, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 7, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7640563712, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 7, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7648493568, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 7, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7650852864, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 7, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7653212160, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 7, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7653900288, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 7, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7661830144, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 7, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7662518272, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 7, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7663206400, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 7, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7671136256, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 7, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7671824384, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 7, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7672512512, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 7, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7673200640, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 7, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7673888768, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 7, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7681818624, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 7, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7689748480, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 7, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7690436608, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 7, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7698366464, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 7, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7700725760, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 7, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7708655616, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 7, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7709343744, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 7, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7717273600, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 7, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7719632896, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 7, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7721992192, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 7, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7729922048, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 7, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7737851904, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 7, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7738540032, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 7, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7746469888, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 7, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7754399744, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 7, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7755087872, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 7, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7763017728, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 7, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7770947584, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 7, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7778877440, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 7, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7786807296, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 7, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7794737152, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 7, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7797096448, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 7, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7805026304, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 7, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7807385600, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 7, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7809744896, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 7, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7812104192, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 7, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7814463488, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 7, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7822393344, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 7, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7823081472, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 7, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7823769600, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 7, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7826128896, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 7, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7834058752, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 7, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7834746880, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 7, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7835435008, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 7, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7843364864, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 7, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7844052992, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 7, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7844741120, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 7, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7845429248, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 7, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7847788544, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 7, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7848476672, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 7, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7856406528, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 7, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7858765824, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 7, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7866695680, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 7, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7869054976, "layers": { "layer.0.self_attn.q_proj": 7, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7871414272, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 7, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7879344128, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 7, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7881703424, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 7, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7882391552, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 7, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7883079680, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 7, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7891009536, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 7, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7898939392, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 7, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7906869248, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 7, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7914799104, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 7, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7917158400, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 7, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7925088256, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 7, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7925776384, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 7, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7928135680, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 7, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7936065536, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 7, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7938424832, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 7, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7940784128, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 7, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7941472256, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 7, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7943831552, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 7, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7944519680, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 7, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7946878976, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 7, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7949238272, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 7, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7949926400, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 7, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7952285696, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 7, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7954644992, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 7, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7957004288, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 7, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7957692416, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 7, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7958380544, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 7, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7959068672, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 7, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7961427968, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 7, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7963787264, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 7, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7966146560, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 7, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7968505856, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 7, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7970865152, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 7, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7973224448, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 7, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7975583744, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 7, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7983513600, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 7, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7985872896, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 7, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7993802752, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 7, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7994490880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 7, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7995179008, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 7, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7995867136, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 7, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7998226432, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 7, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7998914560, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 7, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 7999602688, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 7, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8007532544, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 7, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8008220672, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 7, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8010579968, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 7, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8011268096, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 7, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8013627392, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 7, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8014315520, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 7, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8015003648, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 7, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8015691776, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 7, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8018051072, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 7, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8020410368, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 7, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8022769664, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 7, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8025128960, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 7, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8033058816, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 7, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8035418112, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 7, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8036106240, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 7, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8036794368, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 7, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8044724224, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 7, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8045412352, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 7, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8046100480, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 7, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8048459776, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 7, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8049147904, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 7, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8049836032, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 7, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8050524160, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 7, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8058454016, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 7, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8060813312, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 7, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8068743168, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 7, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8069431296, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 7, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8070119424, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 7, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8078049280, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 7, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8080408576, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 7, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8088338432, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 7, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8089026560, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 7, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8091385856, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 7, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8092073984, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 7, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8094433280, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 7, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8095121408, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 7, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8097480704, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 7, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8105410560, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 7, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8106098688, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 7, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8108457984, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 7, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8109146112, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 7, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8109834240, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 7, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8112193536, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 7, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8112881664, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 7, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8115240960, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 7, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8117600256, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 7, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8125530112, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 7, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8133459968, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 7, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8141389824, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 7, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8149319680, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 7, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8151678976, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 7, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8152367104, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 7, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8154726400, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 7, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8155414528, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 7, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8163344384, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 7, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8171274240, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 7, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8179204096, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 7, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8187133952, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 7, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8187822080, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 8 } }, { "memory": 8195751936, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 8, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 8203681792, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 8, "layer.31.mlp.down_proj": 9 } }, { "memory": 8211611648, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 8, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8219541504, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 8, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8227471360, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 8, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8235401216, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 8, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8236089344, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 8, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8244019200, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 8, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8251949056, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 8, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8259878912, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 8, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8267808768, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 8, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8275738624, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 8, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8276426752, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 8, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8284356608, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 8, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8292286464, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 8, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8300216320, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 8, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8308146176, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 8, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8316076032, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 8, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8324005888, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 8, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8331935744, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 8, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8339865600, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 8, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8347795456, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 8, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8355725312, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 8, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8363655168, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 8, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8371585024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 8, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8379514880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 8, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8387444736, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 8, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8395374592, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 8, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8397733888, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 8, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8405663744, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 8, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8413593600, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 8, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8421523456, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 8, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8429453312, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 8, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8437383168, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 8, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8445313024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 8, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8453242880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 8, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8461172736, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 8, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8463532032, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 8, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8471461888, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 8, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8479391744, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 8, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8481751040, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 8, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8489680896, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 8, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8497610752, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 8, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8505540608, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 8, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8513470464, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 8, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8521400320, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 8, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8529330176, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 8, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8530018304, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 8, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8532377600, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 8, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8534736896, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 8, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8535425024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 8, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8543354880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 8, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8545714176, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 8, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8553644032, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 8, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8561573888, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 8, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8563933184, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 8, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8571863040, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 8, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8579792896, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 8, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8587722752, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 8, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8595652608, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 8, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8598011904, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 8, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8605941760, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 8, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8613871616, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 8, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8614559744, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 8, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8622489600, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 8, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8623177728, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 8, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8623865856, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 8, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8631795712, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 8, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8639725568, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 8, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8647655424, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 8, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8655585280, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 8, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8663515136, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 8, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8664203264, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 8, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8672133120, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 8, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8672821248, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 8, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8673509376, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 8, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8681439232, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 8, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8682127360, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 8, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8682815488, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 8, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8683503616, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 8, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8691433472, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 8, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8699363328, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 8, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8707293184, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 8, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8715223040, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 8, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8717582336, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 8, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8725512192, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 8, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8726200320, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 8, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8728559616, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 8, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8730918912, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 8, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8738848768, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 8, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8741208064, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 8, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8741896192, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 8, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8744255488, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 8, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8752185344, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 8, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8754544640, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 8, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8762474496, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 8, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8770404352, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 8, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8772763648, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 8, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8780693504, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 8, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8781381632, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 8, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8783740928, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 8, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8786100224, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 8, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8786788352, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 8, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8787476480, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 8, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8789835776, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 8, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8790523904, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 8, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8791212032, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 8, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8799141888, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 8, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8807071744, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 8, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8815001600, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 8, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8815689728, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 8, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8818049024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 8, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8820408320, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 8, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8828338176, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 8, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8829026304, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 8, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8836956160, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 8, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8837644288, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 8, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8845574144, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 8, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8847933440, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 8, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8848621568, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 8, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8850980864, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 8, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8853340160, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 8, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8855699456, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 8, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8856387584, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 8, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8858746880, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 8, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8859435008, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 8, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8860123136, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 8, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8862482432, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 8, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8864841728, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 8, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8867201024, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 8, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8867889152, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 8, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8870248448, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 8, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8878178304, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 8, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8886108160, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 8, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8894038016, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 8, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8894726144, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 8, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8897085440, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 8, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8905015296, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 8, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8912945152, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 8, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8913633280, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 8, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8915992576, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 8, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8916680704, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 8, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8917368832, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 8, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8925298688, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 8, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8933228544, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 8, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8933916672, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 8, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8934604800, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 8, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8936964096, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 8, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8939323392, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 8, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8941682688, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 8, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8944041984, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 8, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8951971840, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 8, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8959901696, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 8, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8962260992, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 8, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8962949120, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 8, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8965308416, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 8, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8965996544, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 8, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8966684672, "layers": { "layer.0.self_attn.q_proj": 8, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8969043968, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 8, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8971403264, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 8, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8973762560, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 8, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8976121856, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 8, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8978481152, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 8, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8986411008, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 8, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8988770304, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 8, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8991129600, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 8, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8993488896, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 8, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8994177024, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 8, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8994865152, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 8, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8995553280, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 8, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 8997912576, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 8, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9000271872, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 8, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9000960000, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 8, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9001648128, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 8, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9004007424, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 8, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9011937280, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 8, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9019867136, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 8, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9022226432, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 8, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9024585728, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 8, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9026945024, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 8, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9029304320, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 8, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9031663616, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 8, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9032351744, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 8, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9033039872, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 8, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9033728000, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 8, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9041657856, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 8, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9042345984, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 8, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9043034112, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 8, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9043722240, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 8, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9044410368, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 8, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9046769664, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 8, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9049128960, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 8, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9049817088, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 8, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9057746944, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 8, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9058435072, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 8, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9066364928, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 8, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9074294784, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 8, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9082224640, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 8, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9082912768, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 8, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9085272064, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 8, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9087631360, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 8, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9089990656, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 8, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9092349952, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 8, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9093038080, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 8, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9093726208, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 8, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9094414336, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 8, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9095102464, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 8, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9097461760, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 8, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9105391616, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 8, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9106079744, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 8, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9106767872, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 8, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9107456000, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 8, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9109815296, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 8, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9110503424, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 8, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9111191552, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 8, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9113550848, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 8, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9114238976, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 8, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9122168832, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 8, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9130098688, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 8, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9130786816, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 8, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9133146112, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 8, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9141075968, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 8, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9143435264, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 8, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9144123392, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 9, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9144811520, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 9, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9147170816, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 9, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9155100672, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 9, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9155788800, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 9, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9163718656, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 9, "layer.31.mlp.down_proj": 9 } }, { "memory": 9171648512, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 9, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 9 } }, { "memory": 9172336640, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 9, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 9 } }, { "memory": 9174695936, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 9, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 9 } }, { "memory": 9182625792, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 9, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 9 } }, { "memory": 9190555648, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 9, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 9 } }, { "memory": 9192914944, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 9, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 9 } }, { "memory": 9200844800, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 9 } }, { "memory": 9208774656, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 9, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9216704512, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 9, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9224634368, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 9, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9225322496, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 9, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9233252352, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 9, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9241182208, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 9, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9241870336, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 9, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9249800192, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 9, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9257730048, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 9, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9265659904, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 9, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9266348032, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 9, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9274277888, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 9, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9282207744, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 9, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9290137600, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 9, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9292496896, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 9, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9300426752, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 9, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9308356608, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 9, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9316286464, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 9, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9324216320, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 9, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9326575616, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 9, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9334505472, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 9, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9342435328, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 9, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9350365184, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 9, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9352724480, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 9, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9360654336, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 9, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9363013632, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 9, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9370943488, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 9, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9378873344, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 9, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9386803200, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 9, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9389162496, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 9, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9397092352, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 9, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9405022208, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 9, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9412952064, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 9, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9420881920, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 9, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9428811776, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 9, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9436741632, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 9, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9439100928, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 9, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9447030784, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 9, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9454960640, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 9, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9462890496, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 9, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9463578624, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 9, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9471508480, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 9, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9479438336, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 9, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9487368192, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 9, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9489727488, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 9, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9497657344, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 9, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9505587200, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 9, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9513517056, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 9, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9514205184, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 9, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9516564480, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 9, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9524494336, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 9, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9525182464, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 9, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9533112320, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 9, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9541042176, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 9, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9543401472, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 9, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9551331328, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 9, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9553690624, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 9, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9561620480, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 9, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9569550336, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 9, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9571909632, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 9, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9574268928, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 9, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9574957056, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 9, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9582886912, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 9, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9590816768, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 9, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9591504896, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 9, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9599434752, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 9, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9607364608, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 9, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9608052736, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 9, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9615982592, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 9, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9618341888, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 9, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9619030016, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 9, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9619718144, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 9, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9627648000, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 9, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9635577856, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 9, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9637937152, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 9, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9638625280, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 9, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9646555136, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 9, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9654484992, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 9, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9655173120, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 9, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9657532416, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 9, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9665462272, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 9, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9667821568, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 9, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9670180864, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 9, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9672540160, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 9, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9673228288, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 9, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9681158144, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 9, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9689088000, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 9, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9691447296, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 9, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9693806592, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 9, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9696165888, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 9, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9698525184, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 9, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9700884480, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 9, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9703243776, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 9, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9703931904, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 9, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9704620032, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 9, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9712549888, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 9, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9720479744, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 9, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9722839040, "layers": { "layer.0.self_attn.q_proj": 9, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9725198336, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 9, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9727557632, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 9, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9728245760, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 9, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9736175616, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 9, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9738534912, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 9, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9746464768, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 9, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9748824064, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 9, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9756753920, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 9, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9757442048, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 9, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9758130176, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 9, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9760489472, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 9, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9761177600, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 9, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9761865728, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 9, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9762553856, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 9, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9763241984, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 9, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9771171840, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 9, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9773531136, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 9, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9775890432, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 9, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9778249728, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 9, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9780609024, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 9, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9781297152, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 9, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9781985280, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 9, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9782673408, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 9, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9783361536, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 9, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9784049664, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 9, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9791979520, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 9, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9792667648, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 9, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9793355776, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 9, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9795715072, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 9, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9803644928, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 9, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9804333056, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 9, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9806692352, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 9, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9807380480, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 9, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9809739776, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 9, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9812099072, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 9, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9820028928, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 9, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9827958784, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 9, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9828646912, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 9, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9829335040, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 9, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9830023168, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 9, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9830711296, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 9, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9838641152, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 9, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9839329280, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 9, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9841688576, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 9, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9849618432, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 9, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9850306560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 9, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9852665856, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 9, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9855025152, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 9, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9862955008, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 9, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9870884864, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 9, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9878814720, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 9, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9881174016, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 9, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9883533312, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 9, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9885892608, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 9, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9893822464, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 9, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9894510592, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 9, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9902440448, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 9, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9903128576, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 9, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9903816704, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 9, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9906176000, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 9, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9908535296, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 9, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9909223424, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 9, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9911582720, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 9, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9912270848, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 9, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9914630144, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 9, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9916989440, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 9, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9919348736, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 9, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9921708032, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 9, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9924067328, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 9, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9924755456, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 9, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9925443584, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 9, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9933373440, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 9, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9941303296, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 9, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9941991424, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 9, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9942679552, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 9, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9943367680, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 9, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9944055808, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 9, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9946415104, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 9, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9947103232, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 9, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9949462528, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 9, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9951821824, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 9, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9959751680, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 9, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9960439808, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 9, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9962799104, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 9, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9970728960, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 9, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9978658816, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 9, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9981018112, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 9, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9988947968, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 9, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9991307264, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 9, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9991995392, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 9, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 9992683520, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 9, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10000613376, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 9, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10008543232, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 9, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10009231360, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 9, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10009919488, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 9, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10010607616, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 9, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10012966912, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 9, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10020896768, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 9, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10028826624, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 9, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10029514752, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 9, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10031874048, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 9, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10039803904, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 9, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10040492032, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 9, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10048421888, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 9, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10056351744, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 9, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10064281600, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 9, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10072211456, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 9, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10080141312, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 9, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10088071168, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 9, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10088759296, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 9, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10091118592, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 9, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10091806720, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 9, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10092494848, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 9, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10100424704, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 10, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10108354560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 10 } }, { "memory": 10116284416, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 10, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 11 } }, { "memory": 10124214272, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 10, "layer.31.mlp.down_proj": 11 } }, { "memory": 10132144128, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 10, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10140073984, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 10, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10148003840, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 10, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10155933696, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 10, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10163863552, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 10, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10166222848, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 10, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10174152704, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 10, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10182082560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 10, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10182770688, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 10, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10190700544, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 10, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10193059840, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 10, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10200989696, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 10, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10208919552, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 10, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10216849408, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 10, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10224779264, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 10, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10232709120, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 10, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10240638976, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 10, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10248568832, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 10, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10256498688, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 10, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10264428544, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 10, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10272358400, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 10, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10280288256, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 10, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10288218112, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 10, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10296147968, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 10, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10304077824, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 10, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10312007680, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 10, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10319937536, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 10, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10327867392, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 10, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10335797248, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 10, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10343727104, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 10, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10346086400, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 10, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10354016256, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 10, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10361946112, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 10, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10364305408, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 10, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10372235264, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 10, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10374594560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 10, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10382524416, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 10, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10384883712, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 10, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10392813568, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 10, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10400743424, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 10, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10408673280, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 10, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10416603136, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 10, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10424532992, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 10, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10426892288, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 10, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10434822144, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 10, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10442752000, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 10, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10450681856, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 10, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10458611712, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 10, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10466541568, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 10, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10474471424, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 10, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10482401280, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 10, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10490331136, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 10, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10492690432, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 10, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10493378560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 10, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10501308416, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 10, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10501996544, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 10, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10502684672, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 10, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10505043968, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 10, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10507403264, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 10, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10515333120, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 10, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10523262976, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 10, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10523951104, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 10, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10531880960, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 10, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10539810816, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 10, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10540498944, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 10, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10541187072, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 10, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10543546368, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 10, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10545905664, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 10, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10553835520, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 10, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10556194816, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 10, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10564124672, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 10, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10572054528, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 10, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10572742656, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 10, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10573430784, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 10, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10574118912, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 10, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10582048768, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 10, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10582736896, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 10, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10590666752, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 10, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10598596608, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 10, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10606526464, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 10, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10614456320, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 10, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10622386176, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 10, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10630316032, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 10, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10638245888, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 10, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10638934016, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 10, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10641293312, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 10, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10649223168, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 10, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10649911296, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 10, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10652270592, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 10, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10660200448, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 10, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10668130304, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 10, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10676060160, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 10, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10678419456, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 10, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10680778752, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 10, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10681466880, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 10, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10683826176, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 10, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10686185472, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 10, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10686873600, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 10, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10689232896, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 10, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10691592192, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 10, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10692280320, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 10, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10692968448, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 10, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10700898304, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 10, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10703257600, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 10, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10703945728, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 10, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10704633856, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 10, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10712563712, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 10, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10714923008, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 10, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10717282304, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 10, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10725212160, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 10, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10727571456, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 10, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10735501312, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 10, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10743431168, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 10, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10744119296, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 10, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10744807424, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 10, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10745495552, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 10, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10753425408, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 10, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10754113536, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 10, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10754801664, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 10, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10757160960, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 10, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10759520256, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 10, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10760208384, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 10, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10760896512, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 10, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10768826368, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 10, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10771185664, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 10, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10779115520, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 10, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10779803648, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 10, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10787733504, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 10, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10790092800, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 10, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10790780928, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 10, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10793140224, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 10, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10793828352, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 10, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10794516480, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 10, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10796875776, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 10, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10804805632, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 10, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10812735488, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 10, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10813423616, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 10, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10815782912, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 10, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10818142208, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 10, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10820501504, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 10, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10822860800, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 10, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10825220096, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 10, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10827579392, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 10, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10829938688, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 10, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10830626816, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 10, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10831314944, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 10, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10833674240, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 10, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10834362368, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 10, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10836721664, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 10, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10837409792, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 10, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10845339648, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 10, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10846027776, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 10, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10846715904, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 10, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10847404032, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 10, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10849763328, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 10, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10857693184, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 10, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10860052480, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 10, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10860740608, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 10, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10863099904, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 10, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10865459200, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 10, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10867818496, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 10, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10870177792, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 10, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10870865920, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 10, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10871554048, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 10, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10879483904, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 10, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10887413760, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 10, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10888101888, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 10, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10888790016, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 10, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10896719872, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 10, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10897408000, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 10, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10899767296, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 10, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10907697152, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 10, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10908385280, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 10, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10916315136, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 10, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10918674432, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 10, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10919362560, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 10, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10920050688, "layers": { "layer.0.self_attn.q_proj": 10, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10922409984, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 10, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10924769280, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 10, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10925457408, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 10, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10926145536, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 10, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10928504832, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 10, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10930864128, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 10, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10938793984, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 10, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10941153280, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 10, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10943512576, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 10, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10944200704, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 10, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10946560000, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 10, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10948919296, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 10, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10956849152, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 10, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10959208448, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 10, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10961567744, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 10, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10969497600, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 10, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10977427456, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 10, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10979786752, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 10, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10980474880, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 10, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10982834176, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 10, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10983522304, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 10, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10984210432, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 10, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 10992140288, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 10, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11000070144, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 10, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11002429440, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 10, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11003117568, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 10, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11003805696, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 10, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11006164992, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 10, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11014094848, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 10, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11014782976, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 10, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11022712832, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 10, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11023400960, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 10, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11024089088, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 10, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11024777216, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 10, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11025465344, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 10, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11026153472, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 10, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11026841600, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 10, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11027529728, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 10, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11035459584, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 10, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11037818880, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 10, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11045748736, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 10, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11046436864, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 10, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11054366720, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 10, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11056726016, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 11, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11059085312, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 11 } }, { "memory": 11067015168, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 11, "layer.31.mlp.down_proj": 12 } }, { "memory": 11074945024, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 11, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11082874880, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 11, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11090804736, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 11, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11098734592, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 11, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11106664448, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 11, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11107352576, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 11, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11115282432, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 11, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11123212288, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 11, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11131142144, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 11, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11139072000, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 11, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11147001856, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 11, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11154931712, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 11, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11155619840, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 11, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11157979136, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 11, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11165908992, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 11, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11168268288, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 11, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11176198144, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 11, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11184128000, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 11, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11186487296, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 11, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11188846592, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 11, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11189534720, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 11, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11197464576, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 11, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11205394432, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 11, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11213324288, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 11, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11221254144, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 11, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11229184000, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 11, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11237113856, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 11, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11239473152, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 11, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11247403008, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 11, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11255332864, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 11, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11263262720, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 11, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11271192576, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 11, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11279122432, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 11, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11287052288, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 11, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11294982144, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 11, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11302912000, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 11, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11310841856, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 11, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11318771712, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 11, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11326701568, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 11, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11334631424, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 11, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11342561280, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 11, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11343249408, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 11, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11351179264, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 11, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11359109120, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 11, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11367038976, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 11, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11369398272, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 11, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11370086400, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 11, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11378016256, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 11, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11385946112, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 11, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11386634240, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 11, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11394564096, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 11, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11402493952, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 11, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11404853248, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 11, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11412783104, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 11, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11420712960, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 11, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11428642816, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 11, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11431002112, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 11, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11438931968, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 11, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11441291264, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 11, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11443650560, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 11, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11451580416, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 11, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11459510272, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 11, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11460198400, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 11, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11462557696, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 11, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11464916992, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 11, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11472846848, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 11, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11475206144, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 11, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11483136000, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 11, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11491065856, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 11, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11493425152, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 11, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11501355008, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 11, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11509284864, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 11, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11517214720, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 11, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11519574016, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 11, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11520262144, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 11, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11520950272, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 11, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11528880128, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 11, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11536809984, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 11, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11539169280, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 11, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11539857408, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 11, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11547787264, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 11, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11550146560, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 11, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11552505856, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 11, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11553193984, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 11, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11561123840, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 11, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11563483136, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 11, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11571412992, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 11, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11572101120, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 11, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11574460416, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 11, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11576819712, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 11, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11584749568, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 11, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11585437696, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 11, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11593367552, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 11, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11594055680, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 11, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11596414976, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 11, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11598774272, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 11, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11599462400, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 11, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11601821696, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 11, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11609751552, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 11, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11617681408, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 11, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11620040704, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 11, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11622400000, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 11, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11623088128, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 11, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11631017984, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 11, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11638947840, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 11, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11639635968, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 11, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11640324096, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 11, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11642683392, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 11, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11650613248, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 11, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11651301376, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 11, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11659231232, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 11, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11661590528, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 11, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11662278656, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 11, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11664637952, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 11, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11665326080, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 11, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11667685376, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 11, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11668373504, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 11, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11670732800, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 11, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11671420928, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 11, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11673780224, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 11, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11676139520, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 11, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11676827648, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 11, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11677515776, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 11, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11678203904, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 11, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11680563200, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 11, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11681251328, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 11, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11681939456, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 11, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11682627584, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 11, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11683315712, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 11, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11685675008, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 11, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11693604864, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 11, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11695964160, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 11, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11696652288, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 11, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11699011584, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 11, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11699699712, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 11, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11707629568, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 11, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11708317696, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 11, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11716247552, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 11, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11716935680, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 11, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11719294976, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 11, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11727224832, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 11, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11729584128, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 11, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11731943424, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 11, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11732631552, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 11, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11734990848, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 11, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11735678976, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 11, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11743608832, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 11, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11744296960, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 11, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11744985088, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 11, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11747344384, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 11, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11755274240, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 11, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11757633536, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 11, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11758321664, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 11, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11760680960, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 11, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11761369088, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 11, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11763728384, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 11, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11766087680, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 11, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11774017536, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 11, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11776376832, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 11, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11777064960, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 11, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11777753088, "layers": { "layer.0.self_attn.q_proj": 11, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11780112384, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 11, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11782471680, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 11, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11783159808, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 11, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11785519104, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 11, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11793448960, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 11, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11801378816, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 11, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11809308672, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 11, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11811667968, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 11, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11819597824, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 11, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11820285952, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 11, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11828215808, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 11, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11836145664, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 11, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11838504960, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 11, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11839193088, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 11, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11847122944, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 11, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11847811072, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 11, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11848499200, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 11, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11856429056, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 11, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11858788352, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 11, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11859476480, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 11, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11860164608, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 11, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11862523904, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 11, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11864883200, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 11, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11865571328, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 11, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11866259456, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 11, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11874189312, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 11, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11874877440, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 11, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11877236736, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 11, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11885166592, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 11, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11887525888, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 11, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11888214016, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 11, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11890573312, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 11, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11891261440, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 11, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11899191296, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 11, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11899879424, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 11, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11900567552, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 11, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11908497408, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 11, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11909185536, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 11, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11917115392, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 11, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11917803520, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 11, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11925733376, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 11, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11926421504, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 11, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11928780800, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 11, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11929468928, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 11, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11937398784, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 11, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11945328640, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 11, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11947687936, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 11, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11950047232, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 11, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11950735360, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 11, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11953094656, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 11, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11955453952, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 11, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11956142080, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 11, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11964071936, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 11, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11972001792, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 11, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11972689920, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 11, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11980619776, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 11, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11988549632, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 11, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11989237760, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 11, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 11997167616, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 11, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12005097472, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 11, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12013027328, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 12, "layer.31.mlp.down_proj": 12 } }, { "memory": 12020957184, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 12, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 12 } }, { "memory": 12028887040, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 12, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 12 } }, { "memory": 12036816896, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 12, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 12 } }, { "memory": 12039176192, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 12, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 12 } }, { "memory": 12047106048, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 12, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 12 } }, { "memory": 12047794176, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 12, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 12 } }, { "memory": 12055724032, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 12 } }, { "memory": 12063653888, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 12, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12071583744, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 12, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12079513600, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 12, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12087443456, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 12, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12095373312, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 12, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12103303168, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 12, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12111233024, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 12, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12119162880, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 12, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12127092736, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 12, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12135022592, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 12, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12142952448, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 12, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12143640576, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 12, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12151570432, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 12, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12159500288, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 12, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12167430144, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 12, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12175360000, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 12, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12177719296, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 12, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12185649152, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 12, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12193579008, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 12, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12201508864, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 12, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12209438720, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 12, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12211798016, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 12, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12219727872, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 12, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12227657728, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 12, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12230017024, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 12, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12232376320, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 12, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12240306176, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 12, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12248236032, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 12, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12256165888, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 12, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12258525184, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 12, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12266455040, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 12, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12267143168, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 12, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12269502464, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 12, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12277432320, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 12, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12285362176, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 12, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12286050304, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 12, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12293980160, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 12, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12301910016, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 12, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12302598144, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 12, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12310528000, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 12, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12318457856, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 12, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12320817152, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 12, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12328747008, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 12, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12329435136, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 12, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12337364992, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 12, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12339724288, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 12, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12340412416, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 12, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12348342272, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 12, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12349030400, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 12, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12356960256, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 12, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12357648384, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 12, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12360007680, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 12, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12362366976, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 12, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12370296832, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 12, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12378226688, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 12, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12386156544, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 12, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12394086400, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 12, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12402016256, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 12, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12409946112, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 12, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12417875968, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 12, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12425805824, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 12, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12428165120, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 12, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12430524416, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 12, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12438454272, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 12, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12446384128, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 12, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12447072256, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 12, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12447760384, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 12, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12450119680, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 12, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12458049536, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 12, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12465979392, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 12, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12473909248, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 12, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12481839104, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 12, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12489768960, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 12, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12490457088, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 12, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12492816384, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 12, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12493504512, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 12, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12494192640, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 12, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12502122496, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 12, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12510052352, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 12, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12517982208, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 12, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12518670336, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 12, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12519358464, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 12, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12520046592, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 12, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12522405888, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 12, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12523094016, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 12, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12523782144, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 12, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12526141440, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 12, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12526829568, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 12, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12534759424, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 12, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12537118720, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 12, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12545048576, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 12, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12547407872, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 12, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12555337728, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 12, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12556025856, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 12, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12558385152, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 12, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12560744448, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 12, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12568674304, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 12, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12569362432, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 12, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12571721728, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 12, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12572409856, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 12, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12574769152, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 12, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12577128448, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 12, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12579487744, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 12, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12580175872, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 12, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12580864000, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 12, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12581552128, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 12, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12583911424, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 12, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12584599552, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 12, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12586958848, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 12, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12587646976, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 12, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12590006272, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 12, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12597936128, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 12, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12598624256, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 12, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12606554112, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 12, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12608913408, "layers": { "layer.0.self_attn.q_proj": 12, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12611272704, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 12, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12611960832, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 12, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12612648960, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 12, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12615008256, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 12, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12617367552, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 12, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12618055680, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 12, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12618743808, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 12, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12619431936, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 12, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12627361792, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 12, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12629721088, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 12, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12632080384, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 12, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12632768512, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 12, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12633456640, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 12, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12635815936, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 12, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12636504064, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 12, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12644433920, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 12, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12652363776, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 12, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12660293632, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 12, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12668223488, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 12, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12668911616, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 12, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12671270912, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 12, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12673630208, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 12, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12681560064, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 12, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12682248192, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 12, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12682936320, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 12, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12690866176, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 12, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12698796032, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 12, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12701155328, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 12, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12701843456, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 12, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12704202752, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 12, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12706562048, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 12, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12708921344, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 12, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12716851200, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 12, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12724781056, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 12, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12732710912, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 12, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12733399040, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 12, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12734087168, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 12, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12736446464, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 12, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12738805760, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 12, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12746735616, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 12, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12747423744, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 12, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12748111872, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 12, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12750471168, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 12, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12752830464, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 12, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12755189760, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 12, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12757549056, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 12, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12765478912, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 12, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12773408768, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 12, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12775768064, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 12, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12783697920, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 12, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12784386048, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 12, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12786745344, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 12, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12787433472, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 12, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12788121600, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 12, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12790480896, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 12, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12792840192, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 12, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12793528320, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 12, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12795887616, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 12, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12796575744, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 12, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12804505600, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 12, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12805193728, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 12, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12813123584, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 12, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12815482880, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 12, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12817842176, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 12, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12818530304, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 12, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12826460160, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 12, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12828819456, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 12, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12836749312, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 12, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12837437440, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 12, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12839796736, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 12, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12847726592, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 12, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12855656448, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 12, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12856344576, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 12, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12858703872, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 12, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12861063168, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 12, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12868993024, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 12, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12871352320, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 12, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12872040448, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 12, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12879970304, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 12, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12880658432, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 12, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12888588288, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 12, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12889276416, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 12, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12897206272, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 12, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12897894400, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 12, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12905824256, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 12, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12906512384, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 12, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12907200512, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 12, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12915130368, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 12, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12923060224, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 12, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12930990080, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 12, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12933349376, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 12, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12935708672, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 12, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12938067968, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 12, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12938756096, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 12, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12941115392, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 12, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12941803520, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 12, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12942491648, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 12, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12950421504, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 12, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12951109632, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 12, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12959039488, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 12, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12966969344, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 12, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12969328640, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 13, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12970016768, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 13, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12977946624, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 13, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12985876480, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 13 } }, { "memory": 12993806336, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 13, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13001736192, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 13, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13009666048, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 13, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13012025344, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 13, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13014384640, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 13, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13022314496, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 13, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13030244352, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 13, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13038174208, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 13, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13046104064, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 13, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13048463360, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 13, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13050822656, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 13, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13058752512, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 13, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13066682368, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 13, "layer.31.mlp.down_proj": 14 } }, { "memory": 13074612224, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 13, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13082542080, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 13, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13090471936, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 13, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13092831232, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 13, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13100761088, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 13, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13108690944, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 13, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13111050240, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 13, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13118980096, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 13, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13126909952, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 13, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13134839808, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 13, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13142769664, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 13, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13150699520, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 13, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13151387648, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 13, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13159317504, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 13, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13167247360, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 13, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13175177216, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 13, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13183107072, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 13, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13185466368, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 13, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13193396224, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 13, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13201326080, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 13, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13209255936, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 13, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13217185792, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 13, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13225115648, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 13, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13233045504, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 13, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13240975360, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 13, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13248905216, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 13, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13256835072, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 13, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13264764928, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 13, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13272694784, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 13, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13275054080, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 13, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13282983936, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 13, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13285343232, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 13, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13293273088, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 13, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13301202944, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 13, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13309132800, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 13, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13317062656, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 13, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13317750784, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 13, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13320110080, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 13, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13328039936, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 13, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13330399232, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 13, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13332758528, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 13, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13333446656, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 13, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13341376512, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 13, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13342064640, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 13, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13349994496, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 13, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13357924352, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 13, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13358612480, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 13, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13359300608, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 13, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13361659904, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 13, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13369589760, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 13, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13371949056, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 13, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13379878912, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 13, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13380567040, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 13, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13381255168, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 13, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13381943296, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 13, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13382631424, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 13, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13384990720, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 13, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13387350016, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 13, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13388038144, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 13, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13390397440, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 13, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13392756736, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 13, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13400686592, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 13, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13408616448, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 13, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13416546304, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 13, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13418905600, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 13, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13419593728, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 13, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13427523584, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 13, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13435453440, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 13, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13443383296, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 13, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13445742592, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 13, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13446430720, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 13, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13447118848, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 13, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13449478144, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 13, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13450166272, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 13, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13458096128, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 13, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13460455424, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 13, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13462814720, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 13, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13465174016, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 13, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13467533312, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 13, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13469892608, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 13, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13470580736, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 13, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13471268864, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 13, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13473628160, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 13, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13481558016, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 13, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13489487872, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 13, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13491847168, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 13, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13492535296, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 13, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13494894592, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 13, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13497253888, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 13, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13505183744, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 13, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13505871872, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 13, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13513801728, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 13, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13514489856, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 13, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13516849152, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 13, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13517537280, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 13, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13519896576, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 13, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13520584704, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 13, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13522944000, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 13, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13525303296, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 13, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13525991424, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 13, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13526679552, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 13, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13534609408, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 13, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13535297536, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 13, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13543227392, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 13, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13551157248, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 13, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13559087104, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 13, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13559775232, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 13, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13560463360, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 13, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13562822656, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 13, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13570752512, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 13, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13573111808, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 13, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13581041664, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 13, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13588971520, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 13, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13589659648, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 13, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13590347776, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 13, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13592707072, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 13, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13600636928, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 13, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13608566784, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 13, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13609254912, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 13, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13611614208, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 13, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13613973504, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 13, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13616332800, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 13, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13618692096, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 13, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13626621952, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 13, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13627310080, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 13, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13629669376, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 13, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13632028672, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 13, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13632716800, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 13, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13633404928, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 13, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13635764224, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 13, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13643694080, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 13, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13644382208, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 13, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13646741504, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 13, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13647429632, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 13, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13649788928, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 13, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13650477056, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 13, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13658406912, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 13, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13659095040, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 13, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13659783168, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 13, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13660471296, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 13, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13668401152, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 13, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13670760448, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 13, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13678690304, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 13, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13686620160, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 13, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13688979456, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 13, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13689667584, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 13, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13697597440, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 13, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13705527296, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 13, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13707886592, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 13, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13715816448, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 13, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13716504576, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 13, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13724434432, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 13, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13732364288, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 13, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13734723584, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 13, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13735411712, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 13, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13736099840, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 13, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13738459136, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 13, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13739147264, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 13, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13747077120, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 13, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13749436416, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 13, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13751795712, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 13, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13752483840, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 13, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13753171968, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 13, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13753860096, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 13, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13756219392, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 13, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13756907520, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 13, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13759266816, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 13, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13759954944, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 13, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13760643072, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 13, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13768572928, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 13, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13769261056, "layers": { "layer.0.self_attn.q_proj": 13, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13771620352, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 13, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13773979648, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 13, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13781909504, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 13, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13789839360, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 13, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13792198656, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 13, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13792886784, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 13, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13793574912, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 13, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13794263040, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 13, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13794951168, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 13, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13797310464, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 13, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13805240320, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 13, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13813170176, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 13, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13813858304, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 13, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13821788160, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 13, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13829718016, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 13, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13837647872, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 13, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13840007168, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 13, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13840695296, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 13, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13848625152, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 13, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13856555008, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 13, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13864484864, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 13, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13872414720, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 13, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13874774016, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 13, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13882703872, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 13, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13883392000, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 13, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13891321856, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 13, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13892009984, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 13, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13894369280, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 13, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13895057408, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 13, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13897416704, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 13, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13905346560, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 13, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13906034688, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 13, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13913964544, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 13, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13916323840, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 13, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13924253696, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 13, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13924941824, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 13, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13925629952, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 14, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13933559808, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 14, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13941489664, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 14, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13949419520, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 14, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13957349376, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 14, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13965279232, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 14, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13973209088, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 14, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13981138944, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 14, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13989068800, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 14, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13996998656, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 14, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 13999357952, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 14, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14007287808, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 14 } }, { "memory": 14015217664, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 14, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14015905792, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 14, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14023835648, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 14, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14031765504, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 14, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14039695360, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 14, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14047625216, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 14, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14055555072, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 14, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14063484928, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 14, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14071414784, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 14, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14079344640, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 14, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14087274496, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 14, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14095204352, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 14, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14103134208, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 14, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14105493504, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 14, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14113423360, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 14, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14121353216, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 14, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14129283072, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 14, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14137212928, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 14, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14139572224, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 14, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14147502080, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 14, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14155431936, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 14, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14163361792, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 14, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14164049920, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 14, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14166409216, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 14, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14174339072, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 14, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14182268928, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 14, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14182957056, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 14, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14190886912, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 14, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14193246208, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 14, "layer.31.mlp.down_proj": 15 } }, { "memory": 14201176064, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 14, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14201864192, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 14, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14202552320, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 14, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14210482176, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 14, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14218412032, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 14, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14219100160, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 14, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14219788288, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 14, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14227718144, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 14, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14235648000, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 14, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14236336128, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 14, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14244265984, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 14, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14244954112, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 14, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14245642240, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 14, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14248001536, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 14, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14250360832, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 14, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14258290688, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 14, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14266220544, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 14, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14268579840, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 14, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14276509696, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 14, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14278868992, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 14, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14286798848, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 14, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14294728704, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 14, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14302658560, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 14, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14305017856, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 14, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14305705984, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 14, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14313635840, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 14, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14314323968, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 14, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14316683264, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 14, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14319042560, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 14, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14321401856, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 14, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14322089984, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 14, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14324449280, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 14, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14326808576, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 14, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14329167872, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 14, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14329856000, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 14, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14332215296, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 14, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14334574592, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 14, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14342504448, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 14, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14343192576, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 14, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14345551872, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 14, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14346240000, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 14, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14354169856, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 14, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14362099712, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 14, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14362787840, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 14, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14370717696, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 14, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14378647552, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 14, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14379335680, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 14, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14381694976, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 14, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14382383104, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 14, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14384742400, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 14, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14385430528, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 14, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14393360384, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 14, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14395719680, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 14, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14398078976, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 14, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14398767104, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 14, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14401126400, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 14, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14409056256, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 14, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14411415552, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 14, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14419345408, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 14, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14420033536, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 14, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14427963392, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 14, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14435893248, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 14, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14436581376, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 14, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14437269504, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 14, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14439628800, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 14, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14447558656, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 14, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14455488512, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 14, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14457847808, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 14, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14465777664, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 14, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14466465792, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 14, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14468825088, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 14, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14469513216, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 14, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14470201344, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 14, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14472560640, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 14, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14480490496, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 14, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14488420352, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 14, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14489108480, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 14, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14497038336, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 14, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14499397632, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 14, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14501756928, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 14, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14509686784, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 14, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14517616640, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 14, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14525546496, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 14, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14527905792, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 14, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14530265088, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 14, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14532624384, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 14, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14534983680, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 14, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14537342976, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 14, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14538031104, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 14, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14540390400, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 14, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14548320256, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 14, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14549008384, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 14, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14551367680, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 14, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14553726976, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 14, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14554415104, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 14, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14556774400, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 14, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14564704256, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 14, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14572634112, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 14, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14580563968, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 14, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14581252096, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 14, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14583611392, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 14, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14584299520, "layers": { "layer.0.self_attn.q_proj": 14, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14586658816, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 14, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14594588672, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 14, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14602518528, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 14, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14610448384, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 14, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14611136512, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 14, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14619066368, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 14, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14619754496, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 14, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14627684352, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 14, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14630043648, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 14, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14630731776, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 14, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14633091072, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 14, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14633779200, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 14, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14634467328, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 14, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14636826624, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 14, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14644756480, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 14, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14652686336, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 14, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14653374464, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 14, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14661304320, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 14, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14661992448, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 14, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14662680576, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 14, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14663368704, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 14, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14671298560, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 14, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14671986688, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 14, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14672674816, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 14, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14673362944, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 14, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14675722240, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 14, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14678081536, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 14, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14686011392, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 14, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14686699520, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 14, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14689058816, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 14, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14696988672, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 14, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14699347968, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 14, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14701707264, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 14, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14709637120, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 14, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14710325248, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 14, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14712684544, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 14, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14715043840, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 14, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14715731968, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 14, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14718091264, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 14, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14720450560, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 14, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14721138688, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 14, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14729068544, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 14, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14731427840, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 14, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14732115968, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 14, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14740045824, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 14, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14740733952, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 14, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14743093248, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 14, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14743781376, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 14, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14746140672, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 14, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14748499968, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 14, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14749188096, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 14, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14751547392, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 14, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14759477248, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 14, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14760165376, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 14, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14768095232, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 14, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14776025088, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 14, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14778384384, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 14, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14780743680, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 14, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14783102976, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 14, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14791032832, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 14, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14798962688, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 14, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14799650816, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 14, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14800338944, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 14, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14808268800, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 14, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14808956928, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 14, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14809645056, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 14, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14812004352, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 14, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14819934208, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 14, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14827864064, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 14, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14835793920, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 14, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14843723776, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 14, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14851653632, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 14, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14859583488, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 14, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14867513344, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 14, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14868201472, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 14, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14868889600, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 14, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14871248896, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 14, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14871937024, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 14, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14879866880, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 14, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14880555008, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 14, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14881243136, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 14, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14881931264, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 15, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14882619392, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 15, "layer.31.mlp.down_proj": 15 } }, { "memory": 14890549248, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 15, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 14898479104, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 15, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 14906408960, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 15, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 14914338816, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 15 } }, { "memory": 14922268672, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 15, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14930198528, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 15, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14938128384, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 15, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14946058240, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 15, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14948417536, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 15, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14956347392, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 15, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14957035520, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 15, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14964965376, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 15, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14972895232, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 15, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14980825088, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 15, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14988754944, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 15, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 14996684800, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 15, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15004614656, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 15, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15006973952, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 15, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15014903808, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 15, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15015591936, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 15, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15023521792, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 15, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15031451648, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 15, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15039381504, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 15, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15047311360, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 15, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15055241216, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 15, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15057600512, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 15, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15059959808, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 15, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15062319104, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 15, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15070248960, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 15, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15078178816, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 15, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15086108672, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 15, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15086796800, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 15, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15094726656, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 15, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15097085952, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 15, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15105015808, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 15, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15107375104, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 15, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15115304960, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 15, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15123234816, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 15, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15131164672, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 15, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15131852800, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 15, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15132540928, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 15, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15133229056, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 15, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15133917184, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 15, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15136276480, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 15, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15144206336, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 15, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15152136192, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 15, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15154495488, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 15, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15162425344, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 15, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15163113472, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 15, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15171043328, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 15, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15171731456, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 15, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15174090752, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 15, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15182020608, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 15, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15189950464, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 15, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15192309760, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 15, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15200239616, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 15, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15202598912, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 15, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15203287040, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 15, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15211216896, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 15, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15213576192, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 15, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15215935488, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 15, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15218294784, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 15, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15218982912, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 15, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15219671040, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 15, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15227600896, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 15, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15235530752, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 15, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15237890048, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 15, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15240249344, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 15, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15242608640, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 15, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15244967936, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 15, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15247327232, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 15, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15248015360, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 15, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15250374656, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 15, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15252733952, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 15, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15260663808, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 15, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15261351936, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 15, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15269281792, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 15, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15271641088, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 15, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15279570944, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 15, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15281930240, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 15, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15289860096, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 15, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15290548224, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 15, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15292907520, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 15, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15295266816, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 15, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15297626112, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 15, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15305555968, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 15, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15313485824, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 15, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15315845120, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 15, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15318204416, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 15, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15320563712, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 15, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15328493568, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 15, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15329181696, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 15, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15331540992, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 15, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15339470848, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 15, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15341830144, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 15, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15342518272, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 15, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15344877568, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 15, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15345565696, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 15, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15353495552, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 15, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15354183680, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 15, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15356542976, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 15, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15364472832, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 15, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15372402688, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 15, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15380332544, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 15, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15388262400, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 15, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15388950528, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 15, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15389638656, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 15, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15390326784, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 15, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15392686080, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 15, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15393374208, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 15, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15395733504, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 15, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15398092800, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 15, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15400452096, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 15, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15401140224, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 15, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15401828352, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 15, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15402516480, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 15, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15403204608, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 15, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15411134464, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 15, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15413493760, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 15, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15414181888, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 15, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15414870016, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 15, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15415558144, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 15, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15423488000, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 15, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15424176128, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 15, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15424864256, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 15, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15425552384, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 15, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15433482240, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 15, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15434170368, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 15, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15442100224, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 15, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15442788352, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 15, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15450718208, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 15, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15453077504, "layers": { "layer.0.self_attn.q_proj": 15, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15455436800, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 15, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15456124928, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 15, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15458484224, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 15, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15459172352, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 15, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15459860480, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 15, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15467790336, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 15, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15470149632, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 15, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15478079488, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 15, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15486009344, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 15, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15486697472, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 15, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15489056768, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 15, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15491416064, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 15, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15499345920, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 15, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15500034048, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 15, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15507963904, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 15, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15508652032, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 15, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15511011328, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 15, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15518941184, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 15, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15521300480, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 15, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15523659776, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 15, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15531589632, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 15, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15533948928, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 15, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15541878784, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 15, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15542566912, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 15, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15544926208, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 15, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15545614336, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 15, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15546302464, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 15, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15554232320, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 15, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15562162176, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 15, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15570092032, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 15, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15578021888, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 15, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15585951744, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 15, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15588311040, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 15, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15590670336, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 15, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15591358464, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 15, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15592046592, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 15, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15594405888, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 15, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15595094016, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 15, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15603023872, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 15, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15610953728, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 15, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15611641856, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 15, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15614001152, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 15, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15614689280, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 15, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15615377408, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 15, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15616065536, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 15, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15623995392, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 15, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15626354688, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 15, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15628713984, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 15, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15631073280, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 15, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15633432576, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 15, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15634120704, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 15, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15642050560, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 15, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15649980416, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 15, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15652339712, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 15, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15653027840, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 15, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15660957696, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 15, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15668887552, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 15, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15676817408, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 15, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15679176704, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 15, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15679864832, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 15, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15682224128, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 15, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15690153984, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 15, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15698083840, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 15, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15698771968, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 15, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15706701824, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 15, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15709061120, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 15, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15711420416, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 15, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15713779712, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 15, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15721709568, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 15, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15729639424, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 15, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15737569280, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 15, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15745499136, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 15, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15753428992, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 15, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15754117120, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 15, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15754805248, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 15, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15755493376, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 15, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15763423232, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 15, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15771353088, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 15, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15772041216, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 15, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15779971072, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 15, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15780659200, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 15, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15781347328, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 15, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15789277184, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 15, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15797207040, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 15, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15797895168, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 15, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15805825024, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 15, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15806513152, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 15, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15814443008, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 15, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15822372864, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 15, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15830302720, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 15, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } }, { "memory": 15838232576, "layers": { "layer.0.self_attn.q_proj": 16, "layer.0.self_attn.k_proj": 16, "layer.0.self_attn.v_proj": 16, "layer.0.self_attn.o_proj": 16, "layer.0.mlp.gate_proj": 16, "layer.0.mlp.up_proj": 16, "layer.0.mlp.down_proj": 16, "layer.1.self_attn.q_proj": 16, "layer.1.self_attn.k_proj": 16, "layer.1.self_attn.v_proj": 16, "layer.1.self_attn.o_proj": 16, "layer.1.mlp.gate_proj": 16, "layer.1.mlp.up_proj": 16, "layer.1.mlp.down_proj": 16, "layer.2.self_attn.q_proj": 16, "layer.2.self_attn.k_proj": 16, "layer.2.self_attn.v_proj": 16, "layer.2.self_attn.o_proj": 16, "layer.2.mlp.gate_proj": 16, "layer.2.mlp.up_proj": 16, "layer.2.mlp.down_proj": 16, "layer.3.self_attn.q_proj": 16, "layer.3.self_attn.k_proj": 16, "layer.3.self_attn.v_proj": 16, "layer.3.self_attn.o_proj": 16, "layer.3.mlp.gate_proj": 16, "layer.3.mlp.up_proj": 16, "layer.3.mlp.down_proj": 16, "layer.4.self_attn.q_proj": 16, "layer.4.self_attn.k_proj": 16, "layer.4.self_attn.v_proj": 16, "layer.4.self_attn.o_proj": 16, "layer.4.mlp.gate_proj": 16, "layer.4.mlp.up_proj": 16, "layer.4.mlp.down_proj": 16, "layer.5.self_attn.q_proj": 16, "layer.5.self_attn.k_proj": 16, "layer.5.self_attn.v_proj": 16, "layer.5.self_attn.o_proj": 16, "layer.5.mlp.gate_proj": 16, "layer.5.mlp.up_proj": 16, "layer.5.mlp.down_proj": 16, "layer.6.self_attn.q_proj": 16, "layer.6.self_attn.k_proj": 16, "layer.6.self_attn.v_proj": 16, "layer.6.self_attn.o_proj": 16, "layer.6.mlp.gate_proj": 16, "layer.6.mlp.up_proj": 16, "layer.6.mlp.down_proj": 16, "layer.7.self_attn.q_proj": 16, "layer.7.self_attn.k_proj": 16, "layer.7.self_attn.v_proj": 16, "layer.7.self_attn.o_proj": 16, "layer.7.mlp.gate_proj": 16, "layer.7.mlp.up_proj": 16, "layer.7.mlp.down_proj": 16, "layer.8.self_attn.q_proj": 16, "layer.8.self_attn.k_proj": 16, "layer.8.self_attn.v_proj": 16, "layer.8.self_attn.o_proj": 16, "layer.8.mlp.gate_proj": 16, "layer.8.mlp.up_proj": 16, "layer.8.mlp.down_proj": 16, "layer.9.self_attn.q_proj": 16, "layer.9.self_attn.k_proj": 16, "layer.9.self_attn.v_proj": 16, "layer.9.self_attn.o_proj": 16, "layer.9.mlp.gate_proj": 16, "layer.9.mlp.up_proj": 16, "layer.9.mlp.down_proj": 16, "layer.10.self_attn.q_proj": 16, "layer.10.self_attn.k_proj": 16, "layer.10.self_attn.v_proj": 16, "layer.10.self_attn.o_proj": 16, "layer.10.mlp.gate_proj": 16, "layer.10.mlp.up_proj": 16, "layer.10.mlp.down_proj": 16, "layer.11.self_attn.q_proj": 16, "layer.11.self_attn.k_proj": 16, "layer.11.self_attn.v_proj": 16, "layer.11.self_attn.o_proj": 16, "layer.11.mlp.gate_proj": 16, "layer.11.mlp.up_proj": 16, "layer.11.mlp.down_proj": 16, "layer.12.self_attn.q_proj": 16, "layer.12.self_attn.k_proj": 16, "layer.12.self_attn.v_proj": 16, "layer.12.self_attn.o_proj": 16, "layer.12.mlp.gate_proj": 16, "layer.12.mlp.up_proj": 16, "layer.12.mlp.down_proj": 16, "layer.13.self_attn.q_proj": 16, "layer.13.self_attn.k_proj": 16, "layer.13.self_attn.v_proj": 16, "layer.13.self_attn.o_proj": 16, "layer.13.mlp.gate_proj": 16, "layer.13.mlp.up_proj": 16, "layer.13.mlp.down_proj": 16, "layer.14.self_attn.q_proj": 16, "layer.14.self_attn.k_proj": 16, "layer.14.self_attn.v_proj": 16, "layer.14.self_attn.o_proj": 16, "layer.14.mlp.gate_proj": 16, "layer.14.mlp.up_proj": 16, "layer.14.mlp.down_proj": 16, "layer.15.self_attn.q_proj": 16, "layer.15.self_attn.k_proj": 16, "layer.15.self_attn.v_proj": 16, "layer.15.self_attn.o_proj": 16, "layer.15.mlp.gate_proj": 16, "layer.15.mlp.up_proj": 16, "layer.15.mlp.down_proj": 16, "layer.16.self_attn.q_proj": 16, "layer.16.self_attn.k_proj": 16, "layer.16.self_attn.v_proj": 16, "layer.16.self_attn.o_proj": 16, "layer.16.mlp.gate_proj": 16, "layer.16.mlp.up_proj": 16, "layer.16.mlp.down_proj": 16, "layer.17.self_attn.q_proj": 16, "layer.17.self_attn.k_proj": 16, "layer.17.self_attn.v_proj": 16, "layer.17.self_attn.o_proj": 16, "layer.17.mlp.gate_proj": 16, "layer.17.mlp.up_proj": 16, "layer.17.mlp.down_proj": 16, "layer.18.self_attn.q_proj": 16, "layer.18.self_attn.k_proj": 16, "layer.18.self_attn.v_proj": 16, "layer.18.self_attn.o_proj": 16, "layer.18.mlp.gate_proj": 16, "layer.18.mlp.up_proj": 16, "layer.18.mlp.down_proj": 16, "layer.19.self_attn.q_proj": 16, "layer.19.self_attn.k_proj": 16, "layer.19.self_attn.v_proj": 16, "layer.19.self_attn.o_proj": 16, "layer.19.mlp.gate_proj": 16, "layer.19.mlp.up_proj": 16, "layer.19.mlp.down_proj": 16, "layer.20.self_attn.q_proj": 16, "layer.20.self_attn.k_proj": 16, "layer.20.self_attn.v_proj": 16, "layer.20.self_attn.o_proj": 16, "layer.20.mlp.gate_proj": 16, "layer.20.mlp.up_proj": 16, "layer.20.mlp.down_proj": 16, "layer.21.self_attn.q_proj": 16, "layer.21.self_attn.k_proj": 16, "layer.21.self_attn.v_proj": 16, "layer.21.self_attn.o_proj": 16, "layer.21.mlp.gate_proj": 16, "layer.21.mlp.up_proj": 16, "layer.21.mlp.down_proj": 16, "layer.22.self_attn.q_proj": 16, "layer.22.self_attn.k_proj": 16, "layer.22.self_attn.v_proj": 16, "layer.22.self_attn.o_proj": 16, "layer.22.mlp.gate_proj": 16, "layer.22.mlp.up_proj": 16, "layer.22.mlp.down_proj": 16, "layer.23.self_attn.q_proj": 16, "layer.23.self_attn.k_proj": 16, "layer.23.self_attn.v_proj": 16, "layer.23.self_attn.o_proj": 16, "layer.23.mlp.gate_proj": 16, "layer.23.mlp.up_proj": 16, "layer.23.mlp.down_proj": 16, "layer.24.self_attn.q_proj": 16, "layer.24.self_attn.k_proj": 16, "layer.24.self_attn.v_proj": 16, "layer.24.self_attn.o_proj": 16, "layer.24.mlp.gate_proj": 16, "layer.24.mlp.up_proj": 16, "layer.24.mlp.down_proj": 16, "layer.25.self_attn.q_proj": 16, "layer.25.self_attn.k_proj": 16, "layer.25.self_attn.v_proj": 16, "layer.25.self_attn.o_proj": 16, "layer.25.mlp.gate_proj": 16, "layer.25.mlp.up_proj": 16, "layer.25.mlp.down_proj": 16, "layer.26.self_attn.q_proj": 16, "layer.26.self_attn.k_proj": 16, "layer.26.self_attn.v_proj": 16, "layer.26.self_attn.o_proj": 16, "layer.26.mlp.gate_proj": 16, "layer.26.mlp.up_proj": 16, "layer.26.mlp.down_proj": 16, "layer.27.self_attn.q_proj": 16, "layer.27.self_attn.k_proj": 16, "layer.27.self_attn.v_proj": 16, "layer.27.self_attn.o_proj": 16, "layer.27.mlp.gate_proj": 16, "layer.27.mlp.up_proj": 16, "layer.27.mlp.down_proj": 16, "layer.28.self_attn.q_proj": 16, "layer.28.self_attn.k_proj": 16, "layer.28.self_attn.v_proj": 16, "layer.28.self_attn.o_proj": 16, "layer.28.mlp.gate_proj": 16, "layer.28.mlp.up_proj": 16, "layer.28.mlp.down_proj": 16, "layer.29.self_attn.q_proj": 16, "layer.29.self_attn.k_proj": 16, "layer.29.self_attn.v_proj": 16, "layer.29.self_attn.o_proj": 16, "layer.29.mlp.gate_proj": 16, "layer.29.mlp.up_proj": 16, "layer.29.mlp.down_proj": 16, "layer.30.self_attn.q_proj": 16, "layer.30.self_attn.k_proj": 16, "layer.30.self_attn.v_proj": 16, "layer.30.self_attn.o_proj": 16, "layer.30.mlp.gate_proj": 16, "layer.30.mlp.up_proj": 16, "layer.30.mlp.down_proj": 16, "layer.31.self_attn.q_proj": 16, "layer.31.self_attn.k_proj": 16, "layer.31.self_attn.v_proj": 16, "layer.31.self_attn.o_proj": 16, "layer.31.mlp.gate_proj": 16, "layer.31.mlp.up_proj": 16, "layer.31.mlp.down_proj": 16 } } ]